Introduction: Palette Handling

Any image that is 8 bits per pixel or less must have a palette. The palette is an array that contains the colors for displaying the image. The pixel values are indexes into the array. An image can have its own unique palette or it can share a common fixed palette. In either case, every bitmap in memory that is 8 bits per pixel or less has its own copy of a palette (even if it is a common fixed palette).

One way for a bitmap to have a unique palette is to load an image from a file that is 8 bits per pixel or less. In that case, the file contains a palette for the image, and you can assume that the palette is unique. By default, LEADTOOLS loads the palette that is stored in the file and associates it with the bitmap in memory.

Another way for a bitmap to get a unique palette is to create the palette. For example, if you reduce a 24-bit image to 8 bits per pixel, you can let LEADTOOLS create an optimized palette for the image.

By using a unique, optimized palette, you get the best possible image quality, but there is a drawback. When a display device uses a palette (for example, when it is in 256-color mode), it can only use one palette at a time to display colors. Therefore, if you are displaying more than one image at the same time, the current image may look perfect, while all of the others look bad, because they are mapped to the wrong palette. This is called palette shift.

Using a fixed palette solves these palette-shift problems. The best approach is to specify the fixed palette as a display option. Then, LEADTOOLS remaps each image to the fixed palette when it is displayed, without disturbing the individual image's palette. (When displaying 16-, 24-, or 32-bit images on a device with 256 colors or less, LEADTOOLS always uses the fixed palette.)

Alternatively, you can use a color resolution function to specify a fixed palette for each individual image. But then, when you save the image to a file, the fixed palette is saved in the file, and any unique palette information is lost.

If the image is a 24 or 48-bit bitmap it is possible to specify use of Reference Output Medium Metric (ROMM) BGR Color encoding. This is a wide-primary output-referred extended color-gamut BGR color encoding. This encoding is used by digital still picture imaging systems to manipulate, store, transmit, display and print digital images. Unlike sRGB color encoding, the range of colors is not limited to those colors that can be displayed on a CRT monitor, and unlike e-sRGB, these encodings do not use negative RGB colorimetry co-ordinates.

ROMM color order is supported by the CMP/JPEG/Exif JPEG/JTIF/TIFF CMP file formats. Because there are colors that can be represented in RGB, but not in ROMM as well as colors that can be represented in ROMM that cannot be represented in RGB, if you convert from one to the other you lose the colors that can be represented only in one of the color spaces.